I think you'll get what you want if you drop the colon from the second indexer. The way to think about this is by analogy with python's list slicing semantics.

If I have a list, my_list = [1, 2, 3, 4], then

```
my_list[N]
```

means "give me the Nth element of my_list". One of the nice features of Python's list slicing is that negative indices wrap to the end of lists, so

```
my_list[-1]
```

becomes syntactic sugar for "give me the **last** element of my_list".

Adding a colon to your indexer changes the semantics of what you're asking Python to do. When you write

```
my_list[N:]
```

you're now saying "give me the sub-list of elements drawn from my_list, starting at index N." So for example,

```
my_list = [1, 2, 3, 4]
sub_list = my_list[2:] # sub_list = [3,4]
```

In the particular case where N = -1, you're saying "give me the sub-list of elements drawn from my_list, **starting with the last element of the list**. Thus we get

```
my_list = [1, 2, 3, 4]
sub_list = my_list[-1:] # sub_list = [4]
```

Pandas Series (1D), DataFrame (2D), and Panel (3D) objects extend these semantics in a natural way. Indexing into one of these objects with a scalar value (i.e., without a colon), returns an object of one dimension lower than the original object. Indexing with a slice (i.e. using the colon syntax) returns a filtered-down object of the same dimension as the original object.

In the example above, prices is a 2D DataFrame object. When you index into it with the string 'historical_ratio' you've indexed by a scalar, so you get back a 1D Series, because that's the pandas data type of one dimension lower than your original DataFrame. This is analogous to doing my_list[N] in the simple list case. You then index into the resulting series with the slice [-1:], and because you used a slice instead of a scalar, we get a length-1 Series instead of a value. This is analogous to doing my_list[N:] in the list case.

If instead you index with [-1], pandas will return 1.999622. You can think of this as pandas again "lowering the dimension" of our 1D Series to a "0D" object, which is just a simple scalar.

Hope that helps,

-Scott